Abstract:Shadows in remotely sensed imagery occur when objects totally or partially occlude direct light from a source of illumination, generating great difficulty in land cover interpretation and classification because of the loss of spectral information of shaded pixels. In a mountainous environment with rough terrain, shadows are especially pronounced due to the differentiation of direct illumination between sunny and shady slopes. Topographic correction methods, which are widely used to adjust for differences in solar incidence angles, can partly alleviate the impacts of shadows. However, there are two limitations: one is that the contemporary topographic corrections have little effect on areas that have very low incidence angles and areas that are completely without direct solar illumination (cast shadow); another is that their effectiveness is restricted by the data quality and completeness, spatial resolution, and elevation accuracy of the Digital Elevation Model (DEM) data, which is not currently available in all parts of the world. Thus, noise and errors may be introduced in topographic correction during resampling and geometric registration of the target image. This paper proposes a new approach to restore the radiometric information of mountainous cast shadows using a spectral processing technique called “continuum removal” (CR) without the aid of DEM. The CR-based approach makes full use of the spectral information derived from both the shaded pixels and their neighboring nonshaded pixels of the same land cover type. Several Landsat TM images were used to assess the performance of the proposed method. Results indicated that the proposed method can effectively restore the spectral values of shaded pixels more accurately than the ATCOR_3 correction method, especially for very low incidence angle areas and cast shadows. By comparing data values of shaded pixels with nonshaded pixels (pure reference pixels) of their same class, images processed by the proposed method had the lowest average root mean square error (RMSE) between them in visible, NIR and SWIR bands, followed by the ATCOR_3 correction method and the original image. In addition, the proposed method achieved the best classification accuracy, higher than those from the original test image and the ATCOR_3 corrected image generated using 90 m or 30 m spatial resolution DEM. Therefore, the Continuum Removal method is a better alternative for restoring objects obscured by mountainous shadow when adequate DEM data are unavailable and the quality of DEM cannot satisfy the requirements of topographic correction algorithms.

Shadow Constrained DEM Refinement Based on Differentiable Rendering

Multi-resolution Terrain Rendering Using Summed-Area Tables .

DEM Densification Using Perspective Shape from Shading Through Multispectral Imagery

Real-Time shadow volume algorithm for subdivision surface based models

Efficient Terrain Stochastic Differential Efficient Terrain Stochastic Differential Equations for Multipurpose Digital Elevation Model Restoration

Shadow Detection and Reconstruction of High-Resolution Remote Sensing Images in Mountainous and Hilly Environments

Multi-level Global Optimization Approach for DEM Generation from Aerial Imagery

A Simple Approach to Differentiable Rendering of SDFs

Restoration of Information Obscured by Mountainous Shadows Through Landsat TM/ETM+ Images Without the Use of DEM Data: A New Method

Robust Geometry-Preserving Depth Estimation Using Differentiable Rendering

Extension of Differentiable SAR Renderer for Ground Target Reconstruction from Multiview Images and Shadows

Path-space differentiable rendering of participating media

Enhancement Technology of Terrain Visualization Based on DEM

Diff-DEM: A Diffusion Probabilistic Approach to Digital Elevation Model Void Filling

Differentiable Rendering of Neural SDFs through Reparameterization

Dem Densification Using Sfs With Single Multi-Spectral Satellite Image

Shading aware DSM generation from high resolution multi-view satellite images

Refinement of Monocular Depth Maps via Multi-View Differentiable Rendering

Reconstructing High-Resolution DEMs from 3D Terrain Features Using Conditional Generative Adversarial Networks

Differentiable Shadow Mapping for Efficient Inverse Graphics

Photometric multi-view mesh refinement for high-resolution satellite images